← Back to search

io.github.mdfifty50-boop/agent-guard

mdfifty50-boop Scanned 3d ago

MCP server that detects and prevents infinite agent loops — circuit breakers, pattern detection, and

MCP Registry ↗
C
71.6 / 100

Versions

0.1.0latest
first seen Jun 5, 2026
PermissionsTool SafetyAuthAnnotationsCode QualityStabilitySpecVuln HistoryAuthorTransparencyCommunity

Tools 7

register_agent
annotations: none low

Register an agent for loop monitoring. Configure iteration limits and progress thresholds.

agent_id string max_iterations number progress_threshold number
log_action
annotations: none low

Log an agent action for loop detection. Maintains a rolling window of the last 50 actions and returns early warnings.

agent_id string tool_name string result_preview string
detect_loop
annotations: none low

Check if an agent is stuck in an infinite loop. Analyzes action patterns, uniqueness ratio, and repeated signatures.

agent_id string
set_circuit_breaker
annotations: none low

Configure a circuit breaker to auto-terminate or warn when an action pattern repeats too many times.

agent_id string max_repeats number
check_circuit_breaker
annotations: none low

Check if a proposed action should be allowed or blocked by the circuit breaker. Call this BEFORE executing a tool.

agent_id string proposed_tool string
get_stuck_report
annotations: none low

Get a detailed analysis of why an agent is stuck, including action history, patterns, token waste estimate, and recovery recommendations.

agent_id string
get_health_dashboard
annotations: none low

Get an overview of all monitored agents with their current loop risk scores and status.

Permissions 1

env_vars low
Server uses env_vars capabilities via: process.env

Scan Findings 30

low
Tool 'register_agent' has no annotations annotation_checker · 100%
low
Tool 'log_action' has no annotations annotation_checker · 100%
low
Tool 'detect_loop' has no annotations annotation_checker · 100%
low
Tool 'set_circuit_breaker' has no annotations annotation_checker · 100%
low
Tool 'check_circuit_breaker' has no annotations annotation_checker · 100%
low
Tool 'get_stuck_report' has no annotations annotation_checker · 100%
low
Tool 'get_health_dashboard' has no annotations annotation_checker · 100%
info
Sandbox failed to start for behavioral verification behavioral_verifier · 100%
medium
Vulnerable dependency: @modelcontextprotocol/sdk@1.12.0 (GHSA-345p-7cg4-v4c7) dependency_analyzer · 95%
medium
Vulnerable dependency: @modelcontextprotocol/sdk@1.12.0 (GHSA-46jr-4wx8-v453) dependency_analyzer · 95%
medium
Vulnerable dependency: @modelcontextprotocol/sdk@1.12.0 (GHSA-6prh-2h8m-c8cw) dependency_analyzer · 95%
medium
Vulnerable dependency: @modelcontextprotocol/sdk@1.12.0 (GHSA-6qxp-vccf-f47h) dependency_analyzer · 95%
medium
Vulnerable dependency: @modelcontextprotocol/sdk@1.12.0 (GHSA-8r9q-7v3j-jr4g) dependency_analyzer · 95%
medium
Vulnerable dependency: @modelcontextprotocol/sdk@1.12.0 (GHSA-rvq5-wwqv-78pq) dependency_analyzer · 95%
medium
Vulnerable dependency: @modelcontextprotocol/sdk@1.12.0 (GHSA-w48q-cv73-mx4w) dependency_analyzer · 95%
info
package.json metadata manifest_parser · 100%
info
Tool: register_agent manifest_parser · 70%
info
Tool: log_action manifest_parser · 70%
info
Tool: detect_loop manifest_parser · 70%
info
Tool: set_circuit_breaker manifest_parser · 70%
info
Tool: check_circuit_breaker manifest_parser · 70%
info
Tool: get_stuck_report manifest_parser · 70%
info
Tool: get_health_dashboard manifest_parser · 70%
info
Transport: stdio manifest_parser · 90%
info
Required env vars (1) manifest_parser · 80%
info
Sandbox failed to start for output poisoning scan output_poisoning · 100%
low
Permission: env_vars access detected permission_analyzer · 90%
critical
Tool poisoning in 'check_circuit_breaker': Cross-tool sequencing directive poisoning · 85%
info
SBOM generated: 128 components sbom_generator · 100%
medium
No build provenance detected (SLSA L0) slsa_assessor · 90%